1. Field of the Invention
The invention pertains to a "leads over chip" (LOC) semiconductor die assembly and, more particularly, to a method and apparatus for reducing the stress resulting from lodging of filler particles present in plastic encapsulants between the undersides of the lead frame leads and the active surface of the die and improved lead locking of the leads in position over a portion of the active surface of a semiconductor die assembly.
2. State of the Art
The use of LOC semiconductor die assemblies has become relatively common in the industry in recent years. This style or configuration of semiconductor device replaces a "traditional" lead frame with a central, integral support (commonly called a die-attach tab, paddle, or island) to which the back surface of a semiconductor die is secured, with a lead frame arrangement wherein the dedicated die-attach support is eliminated and at least some of the leads extend over the active surface of the die. The die is then adhered to the lead extensions with an adhesive dielectric layer of some sort disposed between the undersides of the lead extensions and the die. Early examples of LOC assemblies are illustrated in U.S. Pat. No. 4,862,245 to Pashby et al. and U.S. Pat. No. 4,984,059 to Kubota et al. More recent examples of the implementation of LOC technology are disclosed in U.S. Pat. Nos. 5,184,208; 5,252,853; 5,286,679; 5,304,842; and 5,461,255. In instances known to the inventors, LOC assemblies employ large quantities or horizontal cross-sectional areas of adhesive to enhance physical support of the die for handling.
Traditional lead frame die assemblies using a die-attach tab place the inner ends of the lead frame leads in close lateral proximity to the periphery of the active die surface where the bond pads are located, wire bonds then being formed between the lead ends and the bond pads. LOC die assemblies, by their extension of inner lead ends over the die, permit physical support of the die from the leads themselves, permit more diverse (including centralized) placement of the bond pads on the active surface, and permit the use of the leads for heat transfer from the die. However, use of LOC die assemblies in combination with plastic packaging of the LOC die assembly has demonstrated some shortcomings of LOC technology as presently practiced in the art.
One of the shortcomings of the prior art LOC semiconductor die assemblies is that the tape used to bond to the lead fingers of the lead frame does not adequately lock the lead fingers in position for the wire bonding process. At times, the adhesive on the tape is not strong enough to fix or lock the lead fingers in position for wire bonding as the lead fingers pull away from the tape before wire bonding. Alternately, the lead fingers will pull away from the tape after wire bonding of the semiconductor die but before encapsulation of the semiconductor die and frame, either causing shorts between adjacent wire bonds or the wire bonds to pull loose from either the bond pads on the die or lead fingers of the frame. While wire bonding fixtures may be used to attempt to overcome these problems, the fixtures and their use add cost to the fabrication process. Additionally, if large amounts of tape are used to fix the lead fingers in place, the reliability performance of the packaged device will be affected as tape absorbs moisture from the surrounding environment, causing problems during encapsulation and potential corrosion problems.
After wire bonding the semiconductor die to the lead fingers of the lead frame, forming an assembly, the most common manner of forming a plastic package about a die assembly is molding and, more specifically, transfer molding. In this process (and with specific reference to LOC die assemblies), a semiconductor die is suspended by its active surface from the underside of inner lead extensions of a lead frame (typically Cu or Alloy 42) by a tape, screen print or spin-on dielectric adhesive layer. The bond pads of the die and the inner lead ends of the frame are then electrically connected by wire bonds (typically Au, although Al and other metal alloy wires have also been employed) by means known in the art. The resulting LOC die assembly, which may comprise the framework of a dual-in-line package (DIP), zig-zag in-line package (ZIP), small outline j-lead package (SOJ), quad flat pack (QFP), plastic leaded chip carrier (PLCC), surface mount device (SMD) or other plastic package configuration known in the art, is placed in a mold cavity and encapsulated in a thermosetting polymer which, when heated, reacts irreversibly to form a highly cross-linked matrix no longer capable of being re-melted.
The thermosetting polymer generally is comprised of three major components: an epoxy resin, a hardener (including accelerators), and a filler material. Other additives such as flame retardants, mold release agents and colorants are also employed in relatively small amounts.
While many variations of the three major components are known in the art, the focus of the present invention resides in the filler materials employed and its effects on the active die surface and improved lead locking of the lead fingers of the frame.
Filler materials are usually a form of fused silica, although other materials such as calcium carbonates, calcium silicates, talc, mica and clays have been employed for less rigorous applications. Powdered, fused quartz is currently the primary filler used in encapsulants. Fillers provide a number of advantages in comparison to unfilled encapsulants. For example, fillers reinforce the polymer and thus provide additional package strength, enhance thermal conductivity of the package, provide enhanced resistance to thermal shock, and greatly reduce the cost of the polymer in comparison to its unfilled state. Fillers also beneficially reduce the coefficient of thermal expansion (CTE) of the composite material by about fifty percent in comparison to the unfilled polymer, resulting in a CTE much closer to that of the silicon or gallium arsenide die. Filler materials, however, also present some recognized disadvantages, including increasing the stiffness of the plastic package, as well as the moisture permeability of the package.
Another previously unrecognized disadvantage discovered by the inventors herein is the damage to the active die surface resulting from encapsulant filler particles becoming lodged or wedged between the underside of the lead extensions and the active die surface during transfer molding of the plastic package about the die and the inner lead ends of the LOC die assembly. The filler particles, which may literally be jammed in position due to deleterious polymer flow patterns and flow imbalances in the mold cavity during encapsulation, place the active die surface under residual stress at the points of contact of the particles. The particles may then damage the die surface or conductive elements thereon, or immediately thereunder, when the package is further stressed (mechanically, thermally, electrically) during post-encapsulation handling and testing.
While it is possible to employ a lower volume of filler in the encapsulating polymer to reduce potential for filler particle lodging or wedging, a drastic reduction in filler volume raises costs of the polymer to unacceptable levels. More importantly, if the volume of the filler in the encapsulating polymer is reduced, as more polymer is used, the reliability of the encapsulated part is affected as the polymer tends to absorb moisture and is more permeable to moisture, thereby causing a variety of problems for the encapsulated part during encapsulation and subsequent use. Currently available filler technology also imposes certain limitations as to practical beneficial reductions in particle size (currently in the 75 to 125 micron range, with the larger end of the range being easier to achieve with consistency) and in the shape of the filler particles. While it is desirable that particles be of generally spherical shape, it has thus far proven impossible to eliminate non-spherical flakes or chips which, in the wrong orientation, maximize stress on the die surface.
Ongoing advances in design and manufacturing technology provide increasingly thinner conductive, semiconductive and dielectric layers in state-of-the-art dice, and the width and pitch of conductors serving various purposes on the active surface of the die are likewise being continually reduced. The resulting die structures, while robust and reliable for their intended uses, must nonetheless become more stress-susceptible due to the minimal strength provided by the minute widths, depths and spacings of their constituent elements. The integrity of active surface die coats such as silicon dioxide, doped silicon dioxides such as phosphorous silicate glass (PSG) or borophosphorous silicate glass (BPSG), or silicon nitride, may thus be compromised by point stresses applied by filler particles, the result being unanticipated shortening of device life if not immediate, detectable damage or alteration of performance characteristics.
The aforementioned U.S. Pat. No. 4,984,059 to Kubota et al. does incidentally disclose several exemplary LOC arrangements which appear to greatly space the leads over the chip or which do not appear to provide significant areas for filler particle lodging. However, such structures may create fabrication and lead spacing and positioning difficulties.
In addition to solving the problems associated with filler particle lodging and damage, it is desirable to have improved lead locking of the lead fingers of the frame during operations involving the semiconductor die. If the gaps between the lead fingers and the semiconductor die are sealed with an underfill material, the adhesive used to bond the lead fingers to the semiconductor die is more effective in locking the lead fingers in position. Furthermore, the use of an underfill material, in addition to the tape, screen print or spin-on dielectric adhesive layer, provides an additional stabilizing means to immobilize the lead fingers in position, thus reducing or eliminating localized stress failures occurring during the transfer molding process. Previously, improving lead finger locking has been approached from the perspective of improved adhesives and by increasing the flexibility of the lead fingers, rather than sealing the gaps between the leadfingers and the semiconductor die.
From the foregoing, the prior art has neither provided for improved locking of the lead fingers to the semiconductor die, nor recognized the stress phenomenon attendant to transfer molding and the use of filled encapsulants, nor provided an LOC structure which beneficially accommodates this phenomenon.